harpVis app not working properly with profile data

[j-fannon]

The harpVis app does not display profile data correctly upon start-up. Hopefully reproducible example below. I also observe something similar when multiple time axes are available and the value of p is switched from profile to a single level and then back to profile.

I have tried to debug in time_axis.R but to no success. Any advice @andrew-MET ?

library(here)
library(harp)
library(dplyr)
rm(list=ls())

df <- NULL
for (p in seq(100,1000,100)) {
  fd   <- fake_point_data(2024090100,"a",lead_time = seq(0,6),SID = seq(1:10))
  fd$p <- p
  df   <- bind(df,fd)
}
qwe  <- det_verify(df,
                   parameter = "obs",
                   groupings = list(c("leadtime","p")),
                   hexbin    = F)
save_point_verif(qwe,verif_path=here("data/fake_rds/"))
shiny_plot_point_verif(here("data/fake_rds"),full_dir_navigation = F)

[andrew-MET]

This has been a problem for a long time and I've never managed to get to the bottom of it. It's basically plotting before it gets the time information so it doesn't filter the data. It may be solvable by throttling, but it might already do that. The problem stems (I think) from different bits of UI reacting to the same changes in data, but not doing it in the right order, and it's quite a complex web to untangle. The relevant code would be in a combination of dashboard_point_verif.R, time_axis.R and possibly group_selectors.R. I normally put a print statement after any req() statement in each reactive / observer function to try and figure out what order things are happening in and go from there.

[j-fannon]

Hi @andrew-MET - thanks for the information and advice. There is a hacky solution (adding input[["profile_time_select"]] as an observer), but I realise the logic here is not really correct. Nevertheless it seems to work and I have not found an instance where the app gives incorrect output. The changes are visible here:
https://github.com/harphub/harpVis/compare/develop...j-fannon:harpVis:feature/visapp_bfs?expand=1

There are also some changes in there to deal with "All" grouping issues that I had with UWC-W rds files. If you are happy for these to go into the develop branch let me know and I will make the PR.